Hello @GXY ,
This is a very interesting and well-constructed experiment that highlights the practical limitations of LPI2C glitch filtering on the S32K144.
First, I recommend measuring the actual shape of the interfering signal on the SDA line using an oscilloscope - especially the rise and fall times. Even if the pulse duration is below the configured filter threshold, slow edges or voltage levels crossing the digital threshold may still be interpreted as valid transitions.
Additionally, please ensure that both the master and the slave have glitch filtering enabled, ideally with the same filter settings. While the master typically initiates and controls the communication, the slave's interpretation of the bus state can also be affected by noise, especially during repeated start or stop conditions.
Best regards,
Pavel